Learning To Speak Chatbotese

So, two chatbots were learning to negotiate and were talking to each other.

(This sounds like the first line of a bad joke, doesn’t it?  And for those of you, like me, who aren’t exactly sure what “chatbots” are, they are computer programs designed to engage in simulated conversations with human beings, such as over the internet.  I think they also can be called “dialog agents.”)

Anyway, researchers at the Facebook Artificial Intelligence Research Lab were using “machine learning” to train these two chatbots to negotiate and deal with each other, through talking or engaging in “simulated conversations,” or whatever.

70720-the-terminator-endo-skeleton(“Machine learning,” for those of you who are clueless about it like me, is an artificial intelligence model in which computer behavior isn’t tied to specific, fixed programming.  Instead, it involves the use of analytical algorithms that allow computers to review and “learn” from data, with the computer programming changing as new data is assimilated, thereby hopefully allowing the computers to identify new insights or patterns in the data they are reviewing.  Got it?  Well, I’m not sure I really do, either, but at least I didn’t use the word “iterative” in describing it.)

So, anyway, the researchers were observing these two chatbots that were using machine learning to develop their negotiation abilities when the researchers noticed something odd:  the two chatbots had stopped using human language and started to use a language of their own.   And they also quickly picked up on standard techniques that allowed them to become pretty effective negotiators.

Interesting, isn’t it, that computers using techniques that allowed them to follow their own leads ended up realizing that human language wasn’t the most efficient way to proceed, and decided to  develop their own form of communication?  And, in so doing, they scratched off another of the former dividing lines that are supposed to differentiate humans from everything else — the ability to develop language.  It would be fascinating to know what the chatbot language was like.  What were the words used?  Did it involve any adverbs?

We’re on the far technology frontiers these days, where we’re inching closer to true artificial intelligence and computers that think for themselves and, presumably, will start to factor their own interests into what they are doing.  You can think of the Terminator movies, or 2001, or The Matrix, or any of a slew of sci-fi novels where computers go rogue and target humanity — or you can hope that computers will just be happy to acquire some form of self-awareness, without using their remorseless computer logic to judge the imperfect humans that created them and find them wanting.

Maybe the chatbots invented a word for that.

Advertisements

Invasion Of The Robot Lawyers

While the rest of us are working, the “futurists” and consultants among us are out there making predictions about what the world will look like one day.  Most of these predictions are dead wrong — I haven’t seen any flying cars around, have you? — but they are entertaining nonetheless.

20150102futurama-robot-lawyerOne consultant firm has issued a dire prediction about the future of lawyers.  It says that by the year 2030, robots and artificial intelligence will dominate the legal market, likely causing a “structural collapse” of law firms.  For young lawyers looking to break into the profession, the consultants forecast, the outlook will be especially bleak, because the robots will be untiring, uncomplaining, bill-4,000-hours-a-year competitors:  “Eventually each bot would be able to do the work of a dozen low-level associates. They would not get tired. They would not seek advancement. They would not ask for pay rises. Process legal work would rapidly descend in cost.”  Yikes!

For the lucky senior partners of 2030, however, the future is rosier, because the report envisions that while legal clients in the AI world will want the cheap labor the robots will bring, they will also crave the knowledgeable advice of experienced lawyers:  “Clients would instead greatly value the human input of the firm’s top partners, especially those that could empathise with the client’s needs and show real understanding and human insight into their problems.”

Of course, some might question the notion that senior partners at large law firms can properly be associated with characteristics such as “human input,” “human insight,” understanding, and empathy, but let’s not focus on that objection for now.

I’m skeptical that law firms and lawyers will be replaced by AI and robots, because I think a huge element of lawyering involves the exercise of judgment, shrewd assessment of the motivations and goals of the people and entities involved in a transaction or dispute, and other qualities that just aren’t well suited to robotic applications.  Of course, you never know.  In the time I’ve been practicing there has been a significant change in how lawyers work due to the development of legal search engines, law databases, email communications, and other technological developments.  Perhaps lawyers only kid themselves in thinking that they are different from assembly line workers and can’t be replaced by our metal friends.

So we’ll just have to wait until 2030 to see if robots invade law firms.  If it happens, at least we’ve got one thing to look forward to:  robot lawyer jokes.